Interaction of genetic predisposition and
epigenetic factors in the development
of anxiety
Dissertation der Fakultät für Biologie der
Ludwig-Maximilian-Universität München
zur Erlangung des akademischen Grades
Doctor rerum naturalium (Dr. rer. nat.)
Vorgelegt von Patrick Oliver Markt,
Diplom Biologie Universität
Tag der mündlichen Prüfung:
8. Januar 2013
Erster Gutachter:
Prof. Rainer Landgraf
Zweiter Gutachter:
Prof. Gisela Grupe
Angst haben wir alle. Der Unterschied liegt in der Frage wovor.
Frank Thiess († 22. Dezember 1977)
Abstract
It is becoming increasingly clear by current research that the continuum of physiological anxiety up to psychopathology is not merely dependent on genes, but is orchestrated by the interplay of genetic predisposition, gene x environment and epigenetic interactions. To consider this interplay, we here took advantage of the rigid genetic predisposition of a selectively bred mouse model exhibiting high anxiety-related behavior (HAB) and tested whether and how enriched environment, a manipulation of housing conditions, is capable of rescuing the genetically driven high anxiety phenotype via gene x environment and/or epigenetic interactions. Indeed, enriched environment exerts a significant anxiolytic effect on HABs of both sexes indicating for the first time that even a rigid genetic predisposition of high anxiety can be rescued by beneficial environmental stimuli. Thereby, a reduced neophobia and a bigger behavioral repertoire of HABs (e.g. social interactions) have been observed with a stronger anxiolysis in males than in females. The behavioral shift is accompanied by an attenuated release of corticosterone after application of a mild stressor. A hyperreactive hypothalamic-pituitary-adrenal (HPA) axis and amygdala constitute the most common symptoms of anxiety disorders, and decreased corticosterone release seems to entail a reduced release of noradrenaline from locus caeruleus (LC) to the medial prefrontal cortex (mPFC), thereby increasing the top-down control of mPFC on amygdala. This would entail less activation of amygdala and thus HPA axis, a consequence we indeed can observe as decreased neuronal activity flow through the amygdala of enriched housed (EE) compared to standard housed (SE) HABs. We suggest that corticotropin-releasing hormone receptor 1 (Crhr1) is critically involved in this phenomenon since (i) HABs compared to low anxiety-related behavior (LAB) mice exhibit higher Crhr1 mRNA in the basolateral amygdala (BLA), (ii) this overexpression can be significantly reduced when HABs are housed in enriched environment and (iii) a bilateral application of a CRHR1 antagonist in the BLA of SE HABs induced a significant anxiolytic effect. Subsequent pyrosequencing identified that enriched environment increased methylation at a CpG site in the promoter of Crhr1, which is located next to a transcription factor binding site (TFB) of the epigenetic transcription factor Yin Yang 1 (YY1), whose mRNA levels are indeed decreased in EE HABs. In silico analysis identified Nr4a1 and D3Ertd300e as critical co-transcription factors, whereas Nr4a1 seems to be regulated by the quantity of available glucocorticoid receptor (GR) and D3Ertd300e positively regulates YY1. Thus, we hypothesize that reduced corticosterone release decreases the availability and thus binding of corticosterone to GR in the BLA. This, in turn decreases the binding affinity of Nr4a1 to D3Ertd300e, which then cannot positively regulate YY1 to decrease or even prevent methylation at the identified CpG site of Crhr1. This would finally result in a differentially methylated region (DMR) with higher methylation levels in EE HABs, which underlies the observed gene expression differences. The identified DMR might therefore be used as a biomarker for high or pathological anxiety. This hypothesized mechanism highlights the possibility that even a rigid genetic predisposition modeling pathological anxiety might be rescued by an epigenetic process that seems to be triggered by beneficial environmental stimuli, thereby raising the exciting possibility for new treatment strategies, which can be utilized complementary to already existing ones.
Table of contents
List of abbreviations ... 4
1.1 Fear, anxiety and the stress response ... 7
1.2 Maintenance of homeostasis ... 8
1.3 CRH system and the critical role of the amygdala ... 10
1.4 Epigenetics - the missing link in psychopathology? ... 14
1.5 From normal to pathological anxiety... 17
1.6 Limitation of actual treatment strategies for pathological anxiety ... 20
1.7 The beneficial effects of enriched environment ... 21
1.8 High anxiety-related behavior mice - a mouse model of pathological anxiety. ... 24
2 Aims of the thesis ... 27
3 Material and Methods ... 29
3.1 Animals and housing conditions... 29
3.2 Behavioral Testing ... 30
3.3 in silico and molecular analyses ... 37
3.4 Pharmacological manipulation: ... 48
3.5 Statistical analyses ... 51
4. Results ... 53
4.1.1. Effect of EE on anxiety-related behavior ... 53
4.1.2. Effects of EE on exploratory behavior and locomotion ... 58
4.1.3. Effects of EE on coping style and stress reactivity ... 62
4.2 Variability and reproducibility of EE: a meta-analysis of beneficial effects ... 63
4.2.2. Meta-analysis of locomotion ... 66
4.2.3 Meta-analysis of coping style ... 67
4.3 Impact of enrichment on “normal” anxiety-related behavior animals ... 67
4.4.2. Impact of EE on exploratory behavior ... 73
4.3.3 Effect of environmental enrichment on coping style and anhedonia ... 74
4.4 Impact of environmental enrichment on anxiety-related behavior of outbred CD1 mice. ... 75
4.5 Impact of duration on the effects elicited by environmental enrichment ... 76
4.5.2 Impact of prolonged enrichment on coping style ... 82
4.6 Comparison of the impact of environmental enrichment between NABs and HABs receiving either 4 or 10 weeks of EE. ... 85
4.7 Contribution of maternal, pup and adolescent behavior to the anxiolytic effect elicited
by environmental enrichment in HABs. ... 86
4.7.1 Impact of maternal behavior on anxiety-related behavior during enrichment. ... 86
4.7.2 Effect of EE on pup behavior ... 87
4.7.3 Impact of EE on adolescent behavior ... 89
4.8 Identification of candidate genes related to anxiety-related behavior. ... 91
4.8.2 Western Blot ... 93
4.9 Pharmacological validation of CRHR1 via an α-helical antagonist ... 94
4.10 Assessment of Crhr1 promoter methylation by pyrosequencing... 95
4.11 Identification of transcriptional regulators by in silico analysis... 96
4.12 Modulation of behavior by epigenetic drugs ... 98
4.13 Evaluation of transgenerational effects ... 101
5 Discussion and future experiments ... 109
6 References ... 121
7 Acknowledgements ... 134
8 Curriculum vitae ... 135
List of abbreviations
5caC 5-carboxylcytosine
5fC 5-formlycytosine
5hMC 5-hydroxymethylcytosine
5mC 5-methycytosine
ABN arched-back nursing
ACTH adrenocorticotropic hormone
AMP adenosine monophosphate
ANOVA analysis of variance
ANS autonomous nervous system
APS adenosine phosphosulfate
ATP adenosine triphosphate
AVP arginine vasopressin
Β1R β1-adrenoceptor
BCA bicinchoninic acid
BLA basolateral amygdala
BNST bed nucleus of the stria terminalis
cDNA complementary DNA
CA cornu ammonis
CeA central amygdala
CG cingulated cortex
CGi CpG island
CNS central nervous system
CNV copy number variant
CP crossing point
CRH corticotropin-releasing factor (CRF) or hormone Crhr1/2 corticotropin-releasing hormone receptor 1/2
Dbh dopamine beta hydroxylase
DC total distance travelled in the inner zone of open field
DG dentate gyrus
DMR differentially methylated region
DNMT DNA methyltransferase
DNMTi DNA methyltransferase inhibtor
DR dorsal raphe nuclei
DSM-IV Diagnostic and statistical manual IV dNTP deoxynucleoside triphosphate
EC total entries in the inner zone of open field
ECL enhanced chemiluminescence
EE enriched environment
EESE animals tested for transgenerational inheritance ELISA enzyme-linked immunosorbent assay
EP elevated platform
FST forced swim test
GAD generalized anxiety disorder
GxE gene x environment
GR glucocorticoid receptor
HAB high anxiety-related behavior
HC hippocampus
HCA home cage activity
HDAC 1/2 histone deacetylase 1/2 HDACi histone deacetylase inhibtor
HPA axis hypothalamic-pituitary-adrenal axis HPG axis hypothalamic-pituitary-gonadal axis
HRP horseradish peroxidase
HT hypothalamus
ICD10 International classification of diseases 10
IP intraperitoneally
IPTG Isopropyl β-D-1-thiogalactopyranoside
KO knock-out
KWA Kruskal-Wallis ANOVA
LAB low anxiety-related behavior
LB lysogeny broth
LC locus caeruleus (formerly denoted as locus coeruleus) LCe latency to enter the inner zone of open field
LD light dark box
LG licking/grooming
MA medial amygdala
MAT methionine adenosyltransferase
MBP methyl binding protein
MBD1/2 methyl CpG binding domain protein 1/2
MeCP2 methyl CpG binding protein 2
mPFC medial prefrontal cortex
MR mineralocorticoid receptor
MWU Mann Whitney-U test
NA/NE noradrenaline/ norepinephrine NAB normal anxiety-related behavior Npsr1 neuropeptid S receptor 1
NPY neuropeptide Y
NR non-responder
NTS nucleus tractus solitarii
OF open field
PD panic disorder
PLB protein loading buffer
PND postnatal day
NGFI-A nerve growth-inducible factor A
PND postnatal day
POMC proopiomelanocortin
PPi pyrophosphate
PTSD post-traumatic stress disorder
PVN paraventricular nucleus of the hypothalamus OCD obsessive compulsive disorder
OF open field
O/N overnight
qPCR quantitative real-time PCR
RIA radio immunoassay
RNAi RNA interference
RT room temperature
SAH S-adenosylhomocysteine
SAHH S-adenosylhomocysteine hydroxylase
SAM sympatho-adrenergic system
SAMe S-adenosylmethionine
SDS PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis
SE standard environment
SIH stress-induced hyperthermia
SNP single nucleotide polymorphism
SNRI selective noradrenaline reuptake inhibitor
ssDNA single-stranded DNA
SSRI selective serotonin reuptake inhibitor SRT stress reactivity test
T1 body temperature measured during SIH before stress
T2 body temperature measured during SIH after stress
TC percent time spent in the center of open field
TF transcription factor
TFB transcription factor binding site TBST Tris-buffered saline with 0.1% Tween
TDG thymine-DNA-glycosylase
Tet family Ten eleven translocation family of proteins Tmem132d transmembrane protein 132d
TSS transcription start site
TST tail suspension test
UCMS unpredictable chronic mild stress
Ucn urocortin
USV ultrasonic vocalization
VSDI voltage-sensitive dye imaging
WB western blot
1.1 Fear, anxiety and the stress response
Evolution gave rise to at least two distinct behavioral systems, which assess emotion and originate from a common background: fear and anxiety (Flanelly et al, 2007). An increasing body of literature suggests the independence of both systems due to different neurological, hormonal, peptidergic and genetic underpinnings. Fear constitutes a reaction to an immediate threat that inherently endangers the individuals live - like a sudden sound, which might indicate the presence of a predator. These aversive stimuli are unconditioned, i.e. the animal shows an immediate innate “fight-or-flight” response to these discrete and specific stimuli. Contrary to fear, anxiety is a prolonged state of negative emotions to stimuli associated with an elevated potential risk of danger (e.g. open spaces and unfamiliarity) elicited by unavoidable stimuli characterized by the same stimulus features like stress - unpredictability and uncontrollability. Thus, fear represents a brief response to stimuli critical for survival and is mainly mediated by the amygdala, whereas anxiety is a prolonged reaction to a variety of stimuli characterized by uncertainty, which is primarily carried by the bed nucleus of the stria terminalis (BNST) and the hypothalamic-pituitary-adrenal (HPA) axis (Depue et al, 2009).
Though fear and anxiety serve different purposes, both are capable of triggering the body’s stress response by sensing potentially adverse changes in the environment denoted as stress. This in turn, modifies neuronal activity and thus, behavior rapidly and enduringly in response to life-threatening stimuli, achieved through an evolutionary conserved and highly complex modulation of neuronal functioning at several levels of the central nervous system (CNS) (Joels et al, 2009). The pattern and magnitude of this complex stress response depends on “[…] the duration of stress exposure (acute versus chronic), the type of stress (physical versus psychological), the stress context (for example, time of day), the developmental stage of the animal (newborn, adult or aged), the animal’s sex and genetic background” (Joels et al, 2009). The stress response per se comprises two waves of stress mediator actions, which are separated in most instances temporarily and spatially. The acute effects of stress take place within a time frame of seconds to minutes and are carried by monoamines like noradrenaline (NA), dopamine or serotonin as well as neuropeptides like corticotropin-releasing hormone (CRH) and arginine vasopressin (AVP). These rapid stress mediators promote appraisal of the situation, alertness, vigilance and the choice of an optimal strategy to face the situation. These actions quickly subside due to the short availability of their mediators and for this reason, the second wave of stress mediators comprises steroids like cortisol (humans) and CORT (rodents) which act within hours to days primarily upon glucocorticoid receptors (GRs) and thus facilitate the consolidation of information associated with the stressor (i.e. memory) (Joels et al, 2009). These two waves of stress mediators are carried by two different systems working complementary to guarantee an optimal function: the sympatho-adrenomedullary (SAM) system one the one and the HPA axis on the other hand.
1.2 Maintenance of homeostasis
The fast-acting SAM system is predominantly activated by fear and orchestrates the immediate “fight-or-flight” response by secreting monoamines and neuropeptides compared to the HPA axis, which finally causes the secretion of steroids to moderate the adaptive long term response that can’t be controlled by the “fight-or-flight” mechanism (fig. 1). Both mechanisms aim to maintain or reinstate homeostasis during stress.
Every individual maintains its inner environment (e.g. temperature, pH) in an intricate equilibrium denoted as homeostasis. External threats or stimuli (i.e. stressors) aim to compromise this equilibrium every time an animal encounters a (potentially) live-threatening situation that ensues fear or anxiety, which in turn activates the SAM system and HPA-axis. Both neuroendocrine protection systems maintain or reinstate the homeostasis to a new point of equilibrium, which enables the animal to cope optimally with the encountered situation. (Engelmann et al, 2004)
Upon stress, the fast-acting SAM system activates the chromaffin cells of the adrenal medulla via a series of preganglionic sympathetic neurons and paravertebral ganglia (fig. 1, Fig. 1: Stress exposure activates the rapid SAM system (NA) and a prolonged stress response via the HPA axis, which releases a series of neuropeptides entailing the secretion of cortisol (humans) or CORT (rodents) from the adrenal cortex. Blue and red dots depict the sympathetic and parasympathetic nervous system, respectively. adrenocorticotropic hormone (ACTH), noradrenaline (NA), paraventricular nucleus of the hypothalamus (PVN). Picture adopted from Ulrich-Lai and Hermann (2009).
shown in blue; Ulrich-Lai and Hermann, 2009). As a consequence, heart rate, vasoconstriction, respiratory rate and energy mobilization are increased to ensure that the animal is able to escape the situation quickly and actively (Engelmann et al, 2004; Ulrich-Lai and Hermann, 2009).
With a higher latency of ca. 2-3min., the HPA axis activates parvocellular neurons in the paraventricular nucleus of the hypothalamus (PVN), which secrete the hypophysiotrophic hormones CRH and AVP from axon terminals in the Zona externa of the median eminence in the hypothalamo-pituitary portal circulation and further to the adenohypophysis. These releasing hormones in turn, promote the synthesis and processing of proopiomelanocortin (POMC) to adrenocorticotropic hormone (ACTH), which travels via the blood stream to its effector organ - the adrenal cortex. Finally, glucocorticoid hormones, i.e. cortisol (humans) or CORT (rodents) are synthesized and released from the Zona fasciculata of the adrenal cortex into the blood stream (fig. 2) (de Kloet et al, 2005; Engelmann et al, 2004; Ulrich-Lai and Hermann, 2009).
Circulating glucocorticoids then promote the mobilization of stored energy by lipolysis and gluconeogenesis, inhibition of growth and reproductive systems and induce behavioral changes like suppression of feeding, increased arousal, vigilance and cognition and Fig. 2: Stress exposure activates the HPA axis by releasing CRH from the PVN, which in turn secretes ACTH into the blood stream to finally cause a release of cortisol (humans) or corticosterone (rodents) from the adrenal cortex. Picture adapted and modified from Walker et al (2010).
potentiate numerous sympathetically mediated effects, such as peripheral vasoconstriction (Charmandari et al, 2005; Engelmann et al, 2004; Ulrich-Lai and Herrmann, 2009). Moreover, direct innervation of the adrenal cortex by the sympathetic nervous system can regulate corticosteroid release highlighting the largely complementary actions that the SAM system and the HPA-axis have during stress (Ulrich-Lai and Herrmann, 2009). When the animal has successfully escaped the live-threatening situation or the stressor, it is of highest importance to end the body’s stress response and restore homeostasis to the pre-stress point of equilibrium. This is achieved via a negative feedback mechanism, which is tightly regulated by mineralglucocorticoid (MR) and glucocorticoid (GR) receptors. MRs are primarily localized in the hippocampus (HC), lateral septum and medial amygdala (MA), whereas GRs are mainly expressed in the PVN (Joels et al, 2004). Both receptors are expressed in the cytosol during their inactive state, each together with a complex assembly of heat shock proteins including HSP90 and members of the Fkbp5 family. The negative feedback of the HPA axis is achieved through a difference in binding affinity of glucocorticoids with MRs exhibiting a tenfold higher affinity compared to GRs. Under basal conditions, glucocorticoids diffuse passively through the phospholipid membrane to occupy MRs, whereat a stressor drastically increases glucocorticoid availability and thus, entails GR occupancy and transduction of the ligand-activated receptors from the cytosol to the nucleus to bind to glucocorticoid response elements to up- or downregulate the expression of various genes (Charmandari et al, 2005; de Kloet et al, 2005; Pariante and Miller, 2001). Thereby, glucocorticoids regulate their own production via a negative feedback loop, which inhibits the synthesis of ACTH and CRH at the level of the pituitary or HC and hypothalamus (HT), respectively (Raison and Miller, 2003).
To summarize, a stressor challenges the homeostasis of an animal facing a live-threatening situation. This encounter causes fear or anxiety, which in turn leads to the activation of two complementary systems denoted as SAM and HPA axis to adjust the equilibrium to a new set-point to maximize the chance of survival by promoting energy mobilization as well as arousal, vigilance and cognition. After successful escape, the stress response is terminated by a negative feedback mechanism at the level of PVN, HC and HT by utilizing the low affinity of GRs to glucocorticoids. Two members of the stress response, the amygdala and the release of CRH as well as the availability of the associated receptors, play an outstanding role in the regulation of anxiety and in the etiology of psychiatric disorders. 1.3 CRH system and the critical role of the amygdala
The role of CRH in the cause and consequences of anxiety, depression and chronic stress have been extensively documented and thus, may be a main pathway through which the effects of stress can shape brain development (Andrus et al., 2012; Arborelius et al, 1999, Dunn et al, 1990). Chang and Hsu (2004) propose a clear evolutionary trail for the origin of the CRH/CRH receptor system. A coevolution gave rise to diuretic hormone and its associated receptors in insects and CRH/CRH receptors in vertebrates. The latter gained the regulation of the stress response to environmental stimuli early during their evolution and exactly this regulatory mechanism of the CRH system emphasizes its importance in anxiety and may explain its homology among a huge class of species.
Fig. 3: The two different receptors of CRH, denoted as R1 and R2 possess different affinities for the peptides of the CRH family. Figure adapted and modified from Dautzenberg et al (2002).
The CRH system comprises a family of peptides on the one and two types of receptors on the other hand. The CRH family of peptides includes CRH, urocortin (Ucn) as well as Ucn 2 and 3, which are distinct from the CRH and Ucn. Among these peptides, CRH is a 41 amino-acid neuropeptide, which is expressed peripherally (e.g. blood vessels, skin) and centrally (e.g. amygdala) underpinning its critical role in integrating the activity of diverse physiological systems by coordinating the behavioral, neuro-endocrine, and autonomic responses to stress. It acts in the brain through at least two different receptor subtypes, referred to as R1 (Crhr1) and R2 (Crhr2) (Binder and Nemeroff, 2010; Heinrichs et al, 1995; Rotzinger et al, 2010; Vale et al, 1981). These two receptors share approximately 70% identity on the amino acid level but exhibit significant differences at the N-terminal extracellular domain, which might account for their agonist selectivity. Fig. 3 depicts the binding affinities of the two receptor types R1 and R2 for the different peptides of the CRH family. R1 can merely bind CRH (also denoted as CRF) as well as Ucn and seems to be implicated in the stress-producing effects of CRH (Chen et al, 1993; Hsu and Hsueh, 2001; Lewis et al., 2001; Lovenberg et al, 1995; Reyes et al, 2001; Rotzinger et al, 2010; Vaughan et al, 1995). R2 binds all Ucns with significantly higher affinity in relation to CRH indicating that these peptides might be its natural ligands. Moreover, R2 is involved primarily in functions not related to anxiety or depression, e.g. suppression of feeding or decreasing blood pressure (Dautzenberg et al, 2002; Grammatopoulos et al, 2012). The importance of the CRH system is further corroborated by studies that have shown recently that CRH is not only released after stress from axonal terminals of the median eminence but is additionally expressed in neuronal populations of the amygdala (Koob, 2008; Swanson et al, 1983), HC (Chen et al, 2001) and locus caeruleus (LC, earlier referred to as locus coeruleus) (Joels et al, 2009; Valentino and Bockstaele, 2008). CRH acts locally and exerts its neuromodulatory effects within seconds after its release to modify neuronal firing patterns (Aldenhoff et al, 1983; Baram and Hatalski, 1998; Gallagher et al, 2008), gene expression and behavior (Bale et al, 2000; Chen et al, 2006; Coste et al, 2000; Koob, 2008; Muller et al. 2003). Joels et al. (2009) propose the existence of strategic hubs denoted as “hot spots”, where different stress mediator receptors are expressed and networks, involved in different aspects of the stress response, are connected: the prefrontal cortex (PFC), amygdala, HC
and LC constitute such a network and strengthen the concept of spatial convergence of action (Bloom, 1984) (fig. 4).
The presence of monoamine (β1R), neuropeptide (Crhr1/2) and glucocorticoid receptors (MR/GR) within the three major nuclei of the amygdala (CeA, MA, BLA) emphasizes its role in anxiety and stress regulation. The amygdala seems to exhibit two main functions: first, to detect and evaluate the salience, significance, ambiguity and unpredictability of stimuli of biological relevance and to link these stimuli to current estimates of biological value (Davis and Whalen, 2001; Pessoa, 2010; Pessoa and Adolphs, 2011; Whalen, 1998). In other words, the amygdala separates the significant from the mundane by establishing affective significance by highlighting some stimuli so as to receive additional processing by other brain regions, while at the same time other stimuli are deemphasized or discarded (Pessoa, 2010). Neuroimaging studies conducted in humans have confirmed preclinical experiments performed in rodents and non-human primates: “[…] the amygdala responds to negative as well as positively valenced stimuli (Breiter et al., 1996; Somerville et al., 2004; Hennenlotter et al., 2005), suggesting it supports learning about the emotional significance of the environment in general” (Tottenham and Sheridan, 2010). Thus, the amygdala constitutes a brain region to determine the relative safety or danger of a situation, which is especially important during early life when the need to evaluate danger of novel events will be greater (Tottenham et al. 2009a). Indeed, the amygdala has a mostly modulatory role in a wide array of networks and thus, is extensively interconnected with cortical and subcortical structures like the ventral subiculum (vSub), LC, PFC and HC (Pessoa and Adolphs, 2011).
Fig. 4: The receptors MR, GR, Crhr1 and 2 as well as the β1-adrenoceptor for noradrenaline cluster in hot spots of the brain, where at least two of the former receptors are highly expressed to orchestrate different aspects of the stress response including decision making, learning and memory and autonomic and emotional responses. β1-adrenoceptor (β1R), basolateral amygdala (BLA), cornu ammonis 1/3 (CA1/3), dentate gyrus (DG), dorsal raphe nuclei (DR), medial amygdala (MA), nucleus tractus soltarii (NTS). Picture adopted from Joels et al. (2009).
Fig. 5: The amygdala allocates processing resources to stimuli by modulating brain structures required to prioritize particular features of information processing in a given situation. Figure adapted and modified from Belujon and Grace (2011).
Fig. 5 depicts the extensive interconnectedness of the amygdala: the mPFC inhibits activation of BLA under basal conditions, whereas stressful conditions relieve BLA from cortical inhibition via dopaminergic neurotrans-mission (Kröner et al., 2005; Rosenkranz and Grace, 1999). This relieve initiates the HPA cascade in the amygdala, where cells within in the amygdala are quickly activated by stress and participate in the earliest reaction to environmental stressors (Honkaniemi et al., 1992). The BLA sends excitatory inputs into CeA, which in turn activates LC NA release via CRH (Belujon and Grace, 2011). Furthermore, LC and BLA have reciprocal connections by which LC can modulate BLA activity via α- and β-adrenergic receptors to allow the integration of the stress mediators NA and CRH (Belujon and Grace, 2011; Joels et al, 2009). Hippocampally-mediated memory formation can be influenced by the amygdala and the hippocampus per se influences the response of the amygdala when emotional stimuli are encountered (Phelps, 2004). Thereby, mPFC, amygdala and HC coordinate their actions during emotional learning: HC inhibits mPFC in new environmental contexts, followed by a release of the amygdala from mPFC inhibition (Kim and Richardson, 2010; Tottenham and Sheridan, 2010). After cessation of the stressful situation, BLA and LC modify vSub activity, which is thought to participate in the decrease of HPA axis activity (Lowry, 2002). Thus, the amygdala allocates processing resources to stimuli by modulating brain structures required to prioritize particular features of information processing in a given situation (Pessoa and Adolphs, 2011) and is not surprising that “[…] stress-induced changes in the amygdala may have downstream effects on the HPA axis that over time can change the structure and function of later stages in the axis” (Brunson et al., 2001b; Tottenham and Sheridan, 2010). It is now well established that susceptibility to psychiatric disorders is due to the combined effects of genetic, environmental and epigenetic factors. The CRH system, with the amygdala being a brain region of major importance and involvement, may serve as an epigenetic key interface between environmental stressors and the etiology of psychiatric disorders (Binder and Nemeroff, 2010). Genetic predisposition might lead to variation in an individual’s response
Tab. 1: Summary of functions thought to be fulfilled by DNA methylation. Functions of particular interest for this thesis are highlighted in red.
to stressful events and as a consequence, both, the genetic load and epigenetic factors participate in the onset of anxiety disorders or depression.
1.4 Epigenetics - the missing link in psychopathology?
The last two decades of research gradually shifted the view of complex disease etiology from genocentric towards gene x environment (GxE) interactions by recognizing the importance of environmental and epigenetic mechanisms (Jaenisch and Bird, 2003; Rutter, 2006). “DNA is no more considered to be the master blueprint […] operating […] in an ecological vacuum. Rather, DNA outlines the overall adaptive potential of an organism through broadly outlined […] physical and behavior dispositions which serve as building material for the final phenotypic outcome in response to specific environmental stimuli” (Templeton 2006). Indeed, accumulating evidence suggests that epigenetics constitutes one of the main and previously missing links among genetics, environment and disease (Barros and Offenbacher, 2009). We refer to epigenetics as heritable processes that regulate the activity status of at least one gene by molecular factors and processes without altering DNA sequence (Skinner et al, 2010; Svrakic et al, 2010). These processes include DNA methylation, posttranslational modifications of histone tails and RNA interference (RNAi). Though the chronology and grade of interconnectedness of these mechanisms is still a matter of debate, it is increasingly accepted that these processes are not independent of each other (Bossdorf et al, 2008). They finally ensue inaccessibility of genes for the transcription machinery via a condensed chromatin structure (Szyf et al, 2008) enabling organisms to integrate environmental signals into their genome (Murgatroyd et al, 2009). DNA methylation is one major epigenetic research subject (Barros and Offenbacher, 2009) and has been linked inter alia to:
Function Reference(s)
alternative splicing Shukla et al. (2011)
cellular differentiation Illingworth and Bird (2009)
genomic imprinting Sasaki and Matsui (2008)
genomic stability Antequera (2003)
inactivation of alternative promoters Illingworth and Bird (2009)
regulation of gene expression Cedar and Bergmann (2009) Suzuki and Bird (2008)
silencing of molecular parasites Antequera (2003)
transgenerational transmission Jablonka et al. (2009)
X-chromosome inactivation Sasaki and Matsui (2008)
In mammals, DNA methylation predominantly occurs in CpG dinucleotides located within a region denoted as CpG island (CGi). Takai and Jones (2003) defined a CGi as a non-random distribution of methylated CpGs, which encompasses a region of ≥500bp with a
Tab. 2: Comparison of features of CpG islands (CGis) and bulk DNA in mammals. Transcription start site (TSS). Data adapted from Antequera, 2003; Lander et al, 2001. G+C content ≥55% and an observed over expected CpG ratio of ≥65% (tab. 2). This definition has been proven as the most efficient and reliable in silico analysis tool until nowadays (Zhan and Han, 2009).
Feature CGi Bulk DNA
G+C content ≥55% 20-25%
CpG [observed/expected] ≥65% 20-21%
methylated no yes
transcriptionally active yes no
chromatin structure open closed
associated with TSSs yes no
Mazzio and Soliman (2012) summarize the mechanisms that are thought to cause epigenetic silencing of a gene: the enzyme DNA methyltransferase (DNMT) adds methyl groups from the methyl donor S-adenosylmethionine to the 5’ carbon atom of cytosines (fig. 6A). This in turn, attracts methyl binding proteins (MBP) like methyl CpG binding domain protein (MBD) 1, MBD2 and methyl CpG binding protein 2 (MeCP2), which act as docking stations for potent repressor complexes. These complexes comprise co-repressors (e.g. Sin3, N-CoR, Mi-Nu2-NuRD) and histone deacetylases (HDACs, e.g. HDAC1/2), which alter histone stability and nucleosome positioning by controlling modifications of H3 and H4 histone tails. These histones become deacetlyated and histone cores are exchanged by more stable variants (e.g. H2A replaced by H2ABbd) to prevent nucleosomal ejection/displacement. These processes are accompanied by stabilization of the linker histone H1 via the proteins HP1α/β. These proteins tether further silencing elements to tightly crowd methylated DNA to nucleosomes. Finally, lamins position heterochromatin along the nuclear envelope, which causes permanent silencing of a gene (fig. 6B).
Fig. 6: The enzyme DNA methyltransferase (DNMT) transfers a CH3 group from S-adenosylmethionine (SAMe) to cytosines (A), thereby initiating a cascade, which finally silences gene expression (B). Methionine adenosyltransferase (MAT), S-adenosylhomocysteine (SAH), S-S-adenosylhomocysteine hydrolase (SAHH). Pictures adapted and modified from Barros et al. (2009), Foley et al. (2008), Jaenisch et al. (2003).
CGis seem to be located mainly at three different sites: most of them are associated with the promoter region of a gene, rendering them as potent targets for gene regulation. Besides the promoter region, CGis can be found in exons (often denoted as “gene body methylation”; Brenet et al, 2011) as well as in CpG shores (Doi et al, 2009). CpG shores are defined as regions within 2000bp of the TSS without being part of a CGi, explaining the term “shore”.
The process of DNA methylation is highly dynamic and - of highest importance - reversible. The ten eleven translocation (Tet) family of proteins is capable to stepwise oxidize 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), further to 5-formyl-cytosine (5fC) and finally to 5-carboxyl5-formyl-cytosine (5CaC), which is then excised by thymine-DNA-glycosylase (TDG) and replaced by cytosine (He et al, 2011; Inoue et al, 2011). DNA methylation can occur during the whole lifespan of an individual and even exerts influence prenatally and transgenerationally, thereby possibly affecting the health of future generations (Jirtle and Skinner, 2007; for a review refer to Masterpasqua, 2009). The following examples will illustrate the potential and time frame in which DNA methylation can take place: Waterland and Jirtle (2003) found evidence that maternal diet (i.e. a prenatal influence) can alter DNA methylation in the offspring: the agouti allele in mice causes a yellow coat color, overweight and diabetes. If pregnant mothers carrying the agouti allele were provided food supplemented with folic acid (a methyl donor), the agouti allele becomes silenced via DNA methylation generating offspring with brown coat color, which is not overweight and not diabetic. The research group of Meaney provided the maybe most popular evidence that the early postnatal environment and the adjacent post-weaning environment can have a dramatic influence on the phenotype of the offspring and grand-offspring (Champagne and Meaney, 2007). Rodents exhibit natural variations in maternal care (Champagne et al, 2003) seen as high vs. low levels of licking/grooming (LG) and arched-back nursing (ABN) in the first week postpartum. Offspring, which received low compared to high levels of LG and ABN had lower levels of hippocampal GR expression and an increased HPA-response to stress. This seems to be caused by high methylation levels of the transcription factor binding site (TFB) of the nerve growth-inducible factor A (NGFI-A) within the GR gene, thereby sterically hindering binding of NGFI-A and thus, expression of GR. This TFB site is highly methylated already at birth and becomes increasingly demethylated when offspring receives high levels of maternal care, whereas low levels of LG and ABN do not alter methylation causing lower expression of GR and finally an increased reactivity of the HPA-axis (Weaver et al, 2009). It was shown that the phenotype of the offspring can be reversed by manipulating housing conditions of the offspring after weaning. When offspring of high and low LG/ABN mothers was raised in an enriched environment (EE, please refer to section 1.7) or isolation respectively, their phenotype was reversed to low and high LG/ABN respectively. Importantly, the phenotype altered by housing conditions was transmitted to the next generation (Champagne and Meaney, 2007), indicating “[…] that "good" environments […] can ameliorate "bad" epigenomes and "bad" environments (i.e., those provoking fear) can pathologize "good" epigenomes” (Svrakic et al, 2011).
The discovery of epigenetic mechanisms, which allow the transmission of traits to the next generation led to the concept of modern synthesis (Bard, 2011) emphasizing that the environment acts on genes to fine-tune and adapt the organism in the best way possible to the environment it is facing (Bonduriansky et al, 2009). An increasing body of literature corroborates the existence of traits that can be inherited transgenerationally: vinclozolin-induced transgenerational adult-onset disease in rats (Anway et al, 2005), transgenerational promotion of long-term potentiation by altered environment in mice (Arai et al, 2009), transgenerational inheritance of maternal care reversible by housing conditions
(Champagne and Meaney, 2007) and gender bias in multiple sclerosis following epigenetic changes in HLA class III risk haplotypes in humans (Chao et al, 2009).
The aforementioned examples highlight the possibility that DNA methylation can act throughout the whole lifetime and beyond. Fig. 7 depicts factors thought to influence the epigenetic profile and thus, methylation of an individual:
Hatchwell and Greally (2007) suggest at least three categories of factors, which influence the phenotype of an individual: i) epigenetic factors like age, sex or drug use, ii) genetic heterogeneity comprising single nucleotide polymorphisms (SNPs) and copy number variants (CNVs) and iii) non-deterministic factors including stochastic factors (e.g. developmental noise) and environmental influences.
Taken together, DNA methylation can activate or silence genes throughout the whole lifetime and beyond and is able to translate environmental influences into the genome. These features enable DNA methylation to crucially influence the phenotype of an individual to the better or the worse (Belsky et al, 2009), depending on the genetic predisposition and the environment the individual is living in and thus, makes it a promising candidate to identify new mechanisms contributing to the onset of anxiety disorders.
1.5 From normal to pathological anxiety
Mental illnesses, with anxiety disorders being the most common in Europe and the USA, constitute the leading cause of disability worldwide (Andlin-Sobocki et al, 2005; Kessler et al, 2005). A lifetime prevalence of 28.8%, the high comorbidity of anxiety with, amongst others, depression (Bateson et al, 2011), substance abuse and tobacco dependence (Thayer and Kuzawa, 2011; Tottenham and Sheridan, 2010) as well as a high rate of individuals responding only partially or not at all to the prescribed drug treatment might indicate why affected individuals suffer extremely (Bateson et al, 2011; Kessler et al, 2005). Thus, it is not surprising that both disorders rank among the most common and proliferating health problems in the world and cause direct (treatment) and indirect (e.g. absence to work) costs
Fig. 7: At least three different categories of factors contribute to the etiology of complex human diseases: epigenetic factors, genetic heterogeneity and non-deterministic factors. Picture adapted from Hatchwell and Greally (2007).
in the range of hundreds of millions dollar worldwide (Johnston et al, 2009; Wong et al, 2001, 2004). Mental disorders are classified either via the “international classification of diseases” (ICD-10-GM2012) of the WHO or the “diagnostic and statistical manual” (DSM IV) of the American Psychiatric Association. Despite intense characterization, correct classification and subsequent treatment remains challenging due to overlapping symptoms (Svrakic et al., 2011) and patients not responding to treatment - either due to a wrong diagnosis or since they do not respond to the prescribed drug(s). Nowadays, six forms of pathological anxiety are described: social and simple phobia, panic disorder (PD), generalized anxiety disorder (GAD), obsessive compulsive disorders (OCD) and posttraumatic stress disorder (PTSD; ICD-10). The diagnostic and treatment difficulties highlight the multigenic and complex nature of anxiety, which is likely to be shaped by environmentally-driven plasticity at the genomic level. In other words, “a single gene may contribute additively and interchangeably to vulnerability […], but its contribution is neither necessary nor sufficient for manifesting the expression of the phenotype of […]” (Lee et al, 2005) an anxiety disorder.
Affected individuals show one to several behavioral, physiological and (epi-) genetic alterations absent in healthy persons with “normal” or average anxiety: persons suffering from anxiety disorders exhibit a bias to interpret harmless or neutral stimuli as rather dangerous (Kim and Gorman, 2005), a bias to favor a negative association when assessing the emotional quality of a situation (Landgraf, 2001) as well as a hyperreactive HPA-axis (Reul and Holsboer, 2002) and amygdala (Shin and Liberzon, 2010). A hyperreactive HPA-axis causes a prolonged release of glucocorticoids, which might damage brain regions important for the negative feedback mechanism (e.g. HIP) leading to a feed-forward mechanism that drives glucocorticoid synthesis indefinitely under the presence of ongoing stressors (Raison and Miller, 2003). Anxiety per se is protective in many settings, whereas an excessive form promotes disability. Pathological anxiety ensues a series of behavioral manifestations comprising phases of “excessive anxiety and worries” that are consistent over at least 6 months (DSM-IV). Disruption of sleep (hyper- or hyposomnia), weight gain or loss, withdrawal from usual activities, concentration problems, fatigue, feeling tensed or restless represent some, but by far not all, criteria (DSM-IV; Leonardo and Hen, 2006).
Despite intensive research over the last decades, a lot of questions remain merely partially or not answered at all: What are the factors contributing to the onset of pathological anxiety? Is pathological anxiety merely an extreme form of normal anxiety? And what does “normal” mean? Svrakic et al. (2011) propose the term adaption disorders instead of personality disorders since maladaptation might reflect the core deficit. A poor or deviant adaptation to the environment and not extreme behavioral traits (though extreme traits may have interfered with successful adaptation) seem to constitute the gist of the matter. Many researchers agree that extreme behavioral traits per se do not necessarily reflect a maladaption or an evolved dysfunction. These traits are operating functionally if they maximize survival and reproduction within a given environment, even if a mechanism is producing distress or impairs quality of life (Bateson et al, 2011; Sachser et al, 2011). Moreover, the term adaption disorders explicitly include the environment as an important source in the etiology of psychiatric disorders.
The crucial question is which mechanisms might be amenable and/or show a maladaption in pathological anxiety? In particular, evidence is provided that stress facilitates or causes a dysregulation of the endocrine system (Elizalde et. al, 2010; Toth et al, 2008) rendering regulation of the HPA-axis as a promising candidate. Myriads of papers indicate that psychiatric disorders indeed cause a severe dysregulation of the HPA-axis, which seems to be mediated, at least partially, by the CRH/CRH receptor system. CRH is the major
regulator of HPA-axis activity and alterations in CNS CRH-containing neuronal circuits are implicated in the pathopsychology of anxiety and depression (Binder and Nemeroff, 2010; Mathew et al, 2008; Reul and Holsboer, 2002, Risbrough and Stein, 2006). Patients suffering from psychiatric disorders show - in most cases - a hyperreactive HPA-axis likely due to a hyperreactive central CRH system (Reul and Holsboer, 2002).
In addition, patients suffering from anxiety disorders (PTSD, PD, GAD, specific and social phobia) often show a hyperreactive amygdala likely due to a decreased inhibition of mPFC (Shin and Liberzon, 2010) as well as a dysregulation of monoamine systems in limbic structures (Elizalde et. al, 2010; Toth et al, 2008) highlighting amygdala as another potential candidate brain region important for pathological anxiety (Tottenham and Sheridan, 2010). LC and amygdala are closely connected (Itoi and Sugimoto, 2010) and the CRH system innervates both brain regions, which in turn activate the SAM system and CNS noradrenergic production (Valentino et al, 1983). Increased fear and alertness, responses relevant for the fight-or-flight reaction are associated with the release of CRH in limbic brain regions, which is mainly mediated by CRHR1 and is dysregulated in depression and anxiety disorders (Arborelious et al, 1999; Heinrichs et al, 1997; Nemeroff, 2009; Reul and Holsboer, 2002).
Less synapses formed in the hippocampus (Bessa et al, 2009), a smaller hippocampal volume (Gilbertson et al., 2002) and a dysregulated negative feedback of the HPA-axis in patients suffering from anxiety disorders (Reul and Holsboer, 2002) qualify HIP as an additional candidate important for adaption processes and thus, regulation of anxiety. The negative feedback of the HPA-axis (i.e. termination of the stress response) is controlled by MRs and GRs in HIP and indeed, affected persons seem to have lower levels of GRs and a reduced neurogenesis in HIP (Elizalde et. al, 2010; Reul and Holsboer, 2002; Toth et al, 2008). The dysregulation of amygdala, HIP and the CRH system in affected individuals offers the possibility that HPA-axis dysregulation can occur at several stages or levels. Research has begun to identify environmental conditions, which are likely to be associated and/or contribute to the onset of psychiatric disorders. An increasing number of researchers emphasize the existence of genetically influenced individual variations in exposure to risky or protective environments (Jaffe and Price, 2007; Kendler and Baker, 2007). “With respect to depression and anxiety, the key focus is on the fact that environments are not randomly distributed. Social selection means that there needs to be a concern regarding the origins of risk environments as well as focus on their effects” (Rutter, 2010). This hypothesis is strengthened by the fact that an individual’s social position is closely related to the amount of perceived psychosocial stress with a low social position linked to an increased amount of psychosocial stress (Thayer and Kuzawa, 2011). This increase in perceived stress further impacts on blood pressure, stress hormone metabolism and immune function (Bindon et al, 1997; Flinn and England, 1997). Recent animal studies have begun to unravel first mechanisms. Chronic stress exposure in adult mice demethylates the Crh gene within the hypothalamus (Elliott et al, 2010) and highlights the potential of psychosocial stress to induce epigenetic changes linked to pathological physiology and behavior (Thayer and Kuzawa, 2011). Moreover, adverse early life experiences (i.e. traumas) like physical assault or sexual abuse, especially in combination with negative immediate onset-provoking adolescent or adult life experiences (e.g. death of a loved one, job loss) are thought to contribute to the onset of psychiatric disorders (Rutter, 2010). These studies highlight that the life history, i.e. the interaction of environmental and epigenetic factors with the genetic predisposition, is crucial for the onset of a psychiatric disorder. Further evidence comes from twin studies consistently showing that environmental influences account for a substantial proportion of the population variance including anxiety and depression (Plomin et al. 2008).
The interplay of brain regions involved in emotional regulation (e.g. amygdala, HIP, LC, mPFC), neuropeptides (CRH, UCN) and receptors (CRHR1) seems to indicate that chronic stress - in combination with and/or because of the exposure to adverse (i.e. stressful) environments - might cause a dysregulation of the HPA-axis, which causes a gradual shift from normal to pathological anxiety and comorbid depression. Therefore, it is not surprising that the majority of treatment strategies tries to reverse the aforementioned pathological alterations.
1.6 Limitation of actual treatment strategies for pathological anxiety
In general, actual treatment strategies can be divided in two main fields: the prescription of one or a combination of drugs aiming to restore neurotransmitter levels in a specific brain region and non-pharmacological therapies. The latter comprise electroconvulsive therapy, transcranial magnetic stimulation, deep brain stimulation, sleep deprivation and different forms of therapies offered by psychotherapists trying to reverse the maladaptive behavioral repertoire acquired by affected individuals (Fava and Kendler, 2000). Since anxiety and depression show a high level of comorbidity, it is assumed that they share some neurobiological features and thus, many drugs prescribed for depression are also used for the treatment of anxiety disorders (Moehler, 2011).
This highlights the first problem in actual treatment for pathological anxiety. Current research still has not identified exact mechanisms, environmental stimuli or a combination thereof to develop a more effective treatment strategy for pathological anxiety. This hypothesis is corroborated by the estimation that up to 60% of patients suffering from anxiety disorders are resistant or refractory to first-line treatment (Lanouette and Stein, 2010). Nonetheless, affected patients must be taken care of in order to attenuate their suffering. Thereby, benzodiazepines (e.g. Alprazolam, Diazepam, Flurazepam) and selective serotonin reuptake inhibitors (SSRIs, e.g. citalopram, fluoxetine, paroxetine) are the current first-line treatment for most anxiety-disorders (Cryan and Sweeney, 2011; Ravindram and Stein, 2010; Sartori et al, 2011) though tricyclic antidepressants, monoamine oxidase inhibitors and serotonin norepinephrine reuptake inhibitors are also used as treatment options (Black, 2006). All of these drugs have in common that they aim to restore or maintain one or more neurotransmitters in a specific brain region. The second and major problem of these drugs and therefore actual treatment is that they act throughout the whole brain at their target receptors (e.g. GABAA receptors for benzodiazepines) and
not solely in the brain region where it is required. These drugs induce - partially very heavy - side effects, which sometimes cause patients to drop out from treatment or severely impairs their quality of life (though their actual symptoms might be reduced) (personal communication with Dr. Rohrbacher). Drug treatment tries to reestablish brain physiology from pathological back to pre-pathological (i.e. normal or physiological) conditions, whereas psychotherapy follows a different strategy.
Up to now, there are a plethora of different psychotherapies comprising inter alia mindfulness-based therapy (Segal, Williams, & Teasdale, 2002), cognitive behavioral therapy (Black, 2006), psychoeducation (Rummel-Kluge et al, 2009), cognitive therapy, exposure therapy, ritual prevention therapies and psychoanalysis (Black, 2006). Most of these psychotherapeutic methodologies try to identify crucial situations in the life history of an individual, which likely contributed to the onset of pathological anxiety and finally led to a behavioral repertoire, which enables the affected person to circumvent aversive conditions. This is exemplified by the following example: affected persons retreat to their homes instead of pursuing their work - by doing so, they try to circumvent even small problems, which regularly occur during every day work, since even the smallest problem
can cause feelings of helplessness and/or panic due to their bias to interpret neutral stimuli as dangerous (personal communication with Dr. Rohrbacher). Psychotherapies enable the individual to relearn or change behavioral strategies and to adopt techniques to circumvent or attenuate feelings of anxiety and panic (e.g. mindfulness, i.e. focusing on a specific task so extensively and exclusively that the patient has no resources to feel anxious or think about adverse situations).
The third problem complicating the treatment of anxiety disorders is the high comorbidity with other disorders like depression or substance abuse (particularly alcoholism) (Tottenham and Sheridan, 2010). Interestingly, alcoholism, binge drinking and anxiety disorders seem to share molecular underpinnings, namely CRHR1 (Treutlein et al, 2006). Animal studies in alcohol preferring msP rats indicate that ad libitum access to alcohol downregulates Crhr1 in the amygdala and nucleus accumbens (Hansson et al, 2007), whereas a history of alcohol dependence in male Wistar rats causes an upregulation of Crhr1 within the amygdala (Sommer et al, 2008). Both brains regions are involved in the control of emotionality. This highlights the possibility that improved treatment not only improves the quality of life from people suffering from pathological anxiety but that individuals suffering from other disorders may be treated too.
Though it is not clear whether a combinatory treatment of drugs and psychotherapy is superior to monotherapy (Black, 2006), many experts recommend it likely since both treatments pursue a different treatment strategy. Drug treatment tries to establish a pre-pathological brain physiology, whereas psychotherapy changes the way an individual behaves when facing a potential anxiety or panic-provoking situation. Both treatment options aim to improve or reverse the consequences of pathological anxiety and to identify the adverse environmental situations contributing or causing anxiety disorders. Importantly, there is a third possible treatment strategy - to identify beneficial environments, which might prevent or ameliorate pathological anxiety. This strategy is of preventive nature and might be well combined with already existing treatments. In the recent years, positive psychology was paid increasing attention likely belonging to the aforementioned third category since it exactly focuses on the effects of positive environments, situations or behaviors (Proyer et al, 2012). In animal research, EE seems to fulfill the same criteria like positive psychology in human research. Thayer and Kuzuwa (2011) state that “[…] we need to know more about whether EE might have positive effects on biology and health via epigenetic modification […] as an addition to the current focus on strategies to avoid the negative health effects of environmental stress. […] Based upon past work on the epigenetic effects of environmental stressors, one exciting possibility is that the health benefits of EE might also carry across multiple generations through epigenetic inheritance”. This statement clearly emphasizes the potential of EE to identify protective environments and new targets for drug development.
1.7 The beneficial effects of enriched environment
Rosenzweig et al. (1978) were the first researchers who examined the consequences of beneficial housing conditions combining “complex inanimate and social stimulation” on rodent behavior and physiology. They therewith created a paradigm nowadays denoted as EE. It offers laboratory mice an enlarged home environment and provides biological relevant stimuli like group housing, shelter, additional nesting material, climbing structures and deep bedding to facilitate manipulation of their microenvironment. These arrangements create a semi-natural environment with a higher complexity, predictability and controllability. It allows physical exercise to improve motor performance and enhances sensory (e.g. visual; Sale et al, 2004) and cognitive capacities (Olsson and Dahlborn, 2002;
Würbel and Garner, 2007). It is of utmost importance to recognize that the design of EE must be according to the needs of the underlying research subject. There are no universal but just rough guidelines how to design an appropriate EE since the designing process is as unique as the research subject and keeps some fallacies (Würbel and Garner, 2007). Many researchers agree that the plurality of different EEs and utilized mouse strains has led to partially controversial results (Chapillon et al, 1999; Nevison et al, 1999; Van de Weerd et al, 1994), which can be circumvented by a well-chosen and self-designed paradigm (fig. 8; it is hard to imagine that every researcher will use the same paradigm or mouse strain). A well designed EE induces a variety of beneficial effects, which can be attributed to either the “arousal” (Walsh and Cummins, 1975) or “learning and memory” hypothesis (tab. 3; Rosenzweig and Bennett, 1996):
Hypothesis Effect Reference(s)
learning & memory
learning and memory Nithianantharajah and Hannan (2006) neural connectivity Nakamura et al. (1999)
increased neurogenesis Bruel-Jungerman et al. (2005) altered brain region activity van Praag et al. (1999)
arousal
more naturalistic behavioral
pattern Kempermann et al. (2010)
object exploration Renner and Rosenzweig (1986)
increase in locomotor and
exploratory activity Prior and Sachser (1995) anxiolytic and
anti-depressive effects Benaroya-Milshtein et al. (2004); Meshi et al, 2006; Olsson and Dahlborn, 2002 The arousal hypothesis favors the possibility that animals housed in an EE are confronted with higher environmental complexity and often novelty (e.g. changed toys), whereas the learning and memory hypothesis tries to attribute observed changes in brain structures and cellular mechanisms to underlying learning processes. Both hypotheses do not encompass Tab. 3: Beneficial effects elicited after housing rodents in EE. Topics of major interest for this thesis are highlighted in red.
Fig. 8: Stylized picture depicting toys with different color and shape, which may be used to design an EE. Picture adapted and modified from Bengoetxea et al, 2012.
all possible factors nor do they rule out each other and should be viewed complementary to our view. Independent of the hypothesis, EE does not solely induce a variety of beneficial effects, it can also offset many of the negative neurobehavioral and physiological consequences of early life adversity (Bredy et al, 2003, 2004; Francis et al, 2002; Laviola et al., 2004; Morley-Fletcher et al, 2003).
Though EE can induce effects during the whole lifespan of an individual, there seems to be a sensitive period in mammals where animals are more susceptible to environmental influences. Champagne and Curley (2005) suggest the postnatal phase as this period since brain circuits are highly plastic as synaptic connections are elaborated and refined. This phase includes the formation of neuronal circuits mediating anxiety and depression-like behavior in mice and raises the possibility that this period may be critical for setting HPA-axis reactivity (Leonardo and Hen, 2006). In spite of intense studies, researchers were not able to delineate variables (i.e. no specific toy, cage size etc.) contributing to or eliciting beneficial effects, pointing towards EE as an emergent phenomenon strengthening the importance of a well-suited design.
Despite difficulties to identify variables important for the design, researchers have begun to unravel mechanisms likely to contribute to one of the most robust findings of EE - anxiolysis (Sztainberg et al, 2010a). The following examples highlight some of the different mechanisms thought to cause or contribute to the observed anxiolytic effect elicited EE. A study performed by Sztainberg et al. (2010b) indicated that EE decreased Crhr1 mRNA levels in the BLA of C57BL/6 mice compared to controls and might be accounted for the anxiolytic effect. Findings from Okuda et al. (2009) support the potential of EE to influence the amygdala, a brain region of major importance for the regulation of anxiety. They were able to show that progenitor cell proliferation and differentiation are increased in enriched housed C57BL/6 mice compared to controls and suggest that these effects contribute to the anxiolytic effects of their EE procedure.
Besides the amygdala, HIP seems to be another brain region amenable for the effects of EE. Increased hippocampal neurogenesis is of the most robust findings elicited by EE and there is increasing evidence that it may contribute to the observed anxiolytic effects (Revest et al, 2009). The beneficial effects of EE on neurogenesis are thought to be multifactorial comprising inter alia brain-derived neurotropic factor (BDNF; Rossi et al, 2006), which is known to be regulated epigenetically. Indeed, EE increases hippocampal BDNF mRNA expression compared to controls by increasing the trimethylation at lysine 4 of histone 3 (H3K4me3, marker for actively transcribed genes) and reduces lysine 9 and lysine 27 histone 3 trimethylation (H3k9me3 and H3k27me3, respectively; markers for silenced genes) (Kuzumaki et al, 2011). EE is capable of modulating HIP by altering epigenetic modifications highlighting the possibility of EE as a tool to study epigenetic and GxE interactions. Importantly, HIP - like amygdala - is another brain region important for the regulation of emotion
A plethora of studies shows that EE reverses adverse early life or prenatal stress effects inter alia by normalizing a dysregulated HPA-axis (Francis et al, 2002) indicating that EE has the potential to attenuate detrimental effects thought to contribute to pathological anxiety. An even more fascinating and intriguing possibility has been revealed by recent literature. It suggests that EE has the potential to pass its beneficial effects from one generation to the next. Arai et al. (2009) were able to show that two weeks of EE not merely increased the long-term potential (LTP) in the hippocampus of ras-grf 1/2 knock-out (KO) mice via the activation of an alternative cAMP/p38 MAP kinase signaling pathway, in addition increased LTP was passed from EE-housed ras-grf 1/2 KO mothers to their offspring, which has never been housed in an EE. An experiment by Leshem and Schulkin (2011) strengthens the possibility of transgenerational inheritance. To evaluate
whether EE can reverse the effects of prenatal stress, female rats were exposed to stress from PND 27-29 and subsequently split in two groups: one was housed in EE and the other in a standard environment SE until mating. The offspring of prenatally stressed rats housed in SE until mating was further distributed to EE or SE. By doing so, the authors were able to evaluate whether the prenatal effect of EE on mothers on the one and the postnatal effect of EE on the offspring of prenatally stressed mothers on the other hand can reverse the effects of prenatal stress. Indeed, prenatal stress caused a transgenerational effect of adversity in SE offspring of prenatally stressed mothers, which was ameliorated by both, parental and offspring EE. This highlights the exciting possibility of EE to influence the health of future generations by attenuating or rescuing pathological phenotypes.
Taken together, EE raises the possibility to be a valuable tool to study the impact of epigenetic and GxE interactions on pathological anxiety since it exerts significant impact on nearly all brain regions or metabolic processes important in the regulation of emotion - amygdala, HIP and regulation of the HPA-axis - not only in the present, but likely in future generations too. Thus, it is surprising that almost all EE studies performed so far rely on “normal” rodents to identify mechanism contributing to pathological anxiety. This raises several problems: first, normal laboratory animals do not exhibit the pathological genetic predisposition of affected individuals and second, these studies miss the possibility to explore the interaction how environmental stimuli may act on and shape a rigid genetic predisposition, thereby creating a whole new scenario almost completely neglected so far. 1.8 High anxiety-related behavior mice - a mouse model of pathological anxiety. Before considering any model organism, one should pay attention to the criteria a good model organism has to meet in order to produce valid and valuable results. A model organism suited for the study of pathological anxiety must fulfill validity criteria and should meet an acceptable cost-benefit ratio. Mice are easy to breed, have a short-reproductive cycle, are amenable for genetic and environmental manipulations and exhibit low maintenance costs, i.e. exhibit an excellent cost-benefit ratio (Cryan and Holmes, 2005). They possess all prerequisites to detect threatening stimuli - perception of a threat per se, association with a specific context and recalling of the respective memory - and thereby, to experience anxiety (Belzung and Philippot, 2007). Moreover, mice offer remarkable similarities with humans at the molecular, anatomical and physiological level (Leonardo and Hen, 2006) empowering them as good animal models for psychiatric disorders. Clément et al. (2002) propose that an appropriate animal model should meet three validity criteria: the same underlying rationale (construct validity), a close approximation of symptoms including psychopathology (e.g. genetic, neuroendocrine and behavioral concomitants of trait anxiety; face validity) and a reverse of symptoms after pharmacological intervention, i.e. an anxiolytic response after receiving anxiolytics (face validity) (Finn et al, 2003; Gordon and Hen, 2004). It is generally accepted that a complex disorder like depression or certain characteristics thereof can’t be adequately mimicked or modeled in mice (e.g. a mouse is not “depressed” or has suicidal thoughts), which led to the concept of “endophenotypes”. Endophenotypes represent a characteristic or symptom of a disorder that can indeed be modeled in the mouse (e.g. hyposomnia, HPA-axis dysregulation etc.) or reflects a behavioral dimension that is necessary to study psychiatric disorders in a preclinical environment (Cryan et al, 2002). Therefore, a good animal exhibits multiple phenotypes on the one and meets as many validity criteria as possible on the other hand.
Almost all prerequisites are met by a mouse model of pathological anxiety, which has been bred for >45 generations at the Max Planck Institute of Psychiatry in Munich. These mice
were bidirectionally and selectively bred from a genetically heterogeneous CD-1 population for high (HAB), “normal” (NAB) or low (LAB) anxiety-related behavior with percent time spent on open arms as key criterion (fig. 9).
Intrastrain comparisons by selective bidirectional breeding approaches hold the potential to investigate the genetic variability of complex, polygenic traits like anxiety (Sartori et al, 2011; Swallow and Garland, 2005) and thus, to unravel mechanisms contributing to pathological anxiety by focusing on particular traits associated with anxiety disorders (Landgraf and Wigger, 2003), differences in receptor functions likely to be associated with differences in anxiety (Overstreet et al, 2003) or avoidance behavior (Brush, 2003). Selective bidirectional breeding increases the frequency of genes associated with a particular trait by shifting the animals' phenotype bidirectionally from the strain mean (Falconer and Mackay, 1996; Krömer et al, 2005) and clusters it around the extremes of the total spectrum typically observed in an outbred strain (Sartori et al, 2011). HAB mice (HABs) were generated by using exactly this procedure and indeed exhibit a series of endophenotypes closely mimicking pathological characteristics of patients suffering from anxiety disorders and - at the same time - meet a variety of criteria of construct, face and predictive validity.
Converging evidence from more than one behavioral test confirms the face validity of a modeled endophenotype (Cryan et al, 2002). Indeed, the high anxiety trait of HABs compared to NABs and LABs has been confirmed by a reduced time spent in the aversive zone in a multitude of tests including inter alia open field (OF), elevated plus maze (EPM), light dark box (LD) (Kromer et al, 2005; Markt unpublished data, 2009; Markt and Sotnikov, in preparation; Muigg et al, 2009) as well as by an increased aversion to fox odor (Sotnikov, Markt et al, 2011) and an increased number of ultrasonic vocalizations (Kessler, Fig. 9: Breeding course of high (HAB), “normal” (NAB) and low (LAB) anxiety-related behavior mice with percent time spent on open arms a key breeding criterion. All sublines originate from an outbred CD1 population.
